Cerebras and G42 Complete 4 exaFLOP AI Supercomputer and Start the March Towards 8 exaFLOPs

Condor Galaxy takes flight towards the goal of nine interconnected supercomputers with 36 exaFLOPs of AI compute capacity

Cerebras Systems, the pioneer in accelerating generative AI, and G42, the leading UAE-based technology holding group, announced that the Condor Galaxy network of nine interconnected supercomputers has reached its second phase of build out. With the completion of Condor Galaxy 1, Cerebras and G42 have broken ground on the Condor Galaxy 2 (CG-2). CG-2 will be four exaFLOPs and 54 million AI-optimized compute cores, expanding the Condor Galaxy network to a total of eight exaFLOPs and 108 million cores upon completion. This represents an important milestone in Cerebras and G42’s plan to build a 36 exaFLOP constellation of AI supercomputers.

Recommended AI News: Riding on the Generative AI Hype, CDP Needs a New Definition in 2024

“G42’s strategic partnership with Cerebras Systems is paving the way toward our joint vision of creating the world’s largest and fastest AI supercomputers,” said Kiril Evtimov, Group CTO, G42. “We support organizations working with unique, massive, and diverse datasets that require extensive AI training capacity. These use cases span across different sectors including healthcare, energy, investments, and climate studies to name a few.”

“Since we started our collaboration, we successfully launched Jais 13B and 30B bilingual models and supported the release of Med42 Clinical LLM. These early successes illustrate the scale and ease of use of the Condor Galaxy platform. As a result, our rate of iteration from idea, to experiment, to solution has never been faster and has proven to be a game changer for us,” said Dr. Andrew Jackson, EVP, Chief AI Officer at Core42, a G42 company and the national-scale enabler for cloud and generative AI.

Recommended AI News: Integral Ad Science Appoints Kumaresh Singh as Senior Vice President, Data Science

Training large AI models requires a massive amount of compute, vast datasets, and specialized AI expertise. The G42 and Cerebras partnership delivers on all three of these elements. By building and commercializing the largest AI supercomputers, the Condor Galaxy constellation enables simple and easy access to the largest and fastest AI compute in the world. The models trained on the Condor Galaxy have the potential to move humanity forward, making significant discoveries and breakthroughs.

“With CG-1 now complete, we’re already seeing the impact and important contributions that this strategic partnership delivers,” says Andrew Feldman, CEO and co-founder, Cerebras Systems. “In partnership with G42, we are changing the worldwide inventory of compute and using our combined expertise to advance AI work in a powerful way, to quickly and efficiently train leading LLMs.”

Condor Galaxy is designed to enable G42 and its customers to train large, ground-breaking models quickly and easily, thereby accelerating the pace of innovation. The Cerebras-G42 strategic partnership has already advanced state-of-the-art AI models in healthcare with the introduction of Med42, the leading generative AI medical assistant, as well as brought Arabic Language chat to more than 400 million Arabic speakers through the introduction of Jais 30B, the premier Arabic language LLM. Pioneering work on CG-1 has also included AI-based climate studies as well as pioneering work in high-performance computing.

Recommended AI News: XConn Technologies Announces Partnership with Liqid

In addition to CG-1 and CG-2, Cerebras and G42 previously announced their plans to deploy seven additional supercomputers — CG-3 through CG-9 — bringing the total compute power of the Condor Galaxy to 36 exaFLOPs. This entails deploying 576 Cerebras CS-2 systems and feeding the cluster with more than 654,000 AMD CPU cores. With 36 exaFLOPs in total, this unprecedented constellation of supercomputers will revolutionize the advancement of AI globally.

[To share your insights with us, please write to sghosh@martechseries.com]